Agentic
Like-for-like- Claude Haiku 4.5
- 27.0
- Supported · #142/152
- Grok 4.20
- 26.7
- Supported · #145/152
- Basis
- BenchAlign lane · 2 vs 4 public rows
- Reading
- Practical tie
Claude Haiku 4.5 vs Grok 4.20
Updated September 10, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
Grok 4.20 has the higher public score estimate, 67.13 versus 52.79, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
6 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Share or export
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Coding work
Code generation, repair, and software-engineering tasks
Grok 4.20
Grok 4.20 leads on the public coding lane, 28.2 to 27.2, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Long documents
Prompts that approach the documented context limit
Grok 4.20
Grok 4.20 has the larger documented context window.
Confidence: documented
Chat turn cost
1K fresh input + 500 output tokens
Claude Haiku 4.5
Claude Haiku 4.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Show secondary and unsupported calls
Repository review cost
50K fresh input + 3K output tokens
Claude Haiku 4.5
Claude Haiku 4.5 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Agentic work
Tool use, computer use, and multi-step task completion
No clear pick
The like-for-like agentic result is a practical tie on the public lane (within 0.5 points).
Confidence: limited
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. Claude Haiku 4.5 does not fit this workload in one request. Grok 4.20 has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallback
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
4 Claude Haiku 4.5 only
6 Shared
17 Grok 4.20 only
- Shared results
- 6
- Claude Haiku 4.5 only
- 4
- Grok 4.20 only
- 17
- Like-for-like categories
- 2 / 8
1 category rests on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
Coding
Like-for-like- Claude Haiku 4.5
- 27.2
- Supported · #144/151
- Grok 4.20
- 28.2
- Supported · #141/151
- Basis
- BenchAlign lane · 4 vs 6 public rows
- Reading
- Grok 4.20 leads · intervals overlap
Knowledge
Directional only- Claude Haiku 4.5
- 44.2
- Estimated · #115/183
- Grok 4.20
- 49.4
- Supported · #85/183
- Basis
- BenchAlign lane · 2 vs 6 public rows
- Reading
- Directional only
Reasoning
Not comparable- Claude Haiku 4.5
- Not ranked
- Grok 4.20
- 34.2
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
Math
Not comparable- Claude Haiku 4.5
- 28.9
- Unranked · 2 rankable rows
- Grok 4.20
- Not ranked
- Basis
- Provisional lane · 2 vs 0 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Claude Haiku 4.5
- Not ranked
- Grok 4.20
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Claude Haiku 4.5
- Not ranked
- Grok 4.20
- 34.6
- #43/48
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Claude Haiku 4.5
- Not ranked
- Grok 4.20
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Claude Haiku 4.5 | Grok 4.20 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 27.0Supported · #142/152 | 26.7Supported · #145/152 | Like-for-likeBenchAlign lane · 2 vs 4 public rows | Practical tie |
| Coding | 27.2Supported · #144/151 | 28.2Supported · #141/151 | Like-for-likeBenchAlign lane · 4 vs 6 public rows | Grok 4.20 leads · intervals overlap |
| Knowledge | 44.2Estimated · #115/183 | 49.4Supported · #85/183 | Directional onlyBenchAlign lane · 2 vs 6 public rows | Directional only |
| Reasoning | Not ranked | 34.2Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Math | 28.9Unranked · 2 rankable rows | Not ranked | Not comparableProvisional lane · 2 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | Not ranked | 34.6#43/48 | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Instruction following | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
LiveCodeBench (Vals)
Coding
MMLU-Pro (Vals)
Knowledge
SWE-bench Verified
Coding
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- Claude Haiku 4.5
- $0.0035
- Fits in one request
- Grok 4.20
- $0.005
- Fits in one request
Claude Haiku 4.5 has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Claude Haiku 4.5
- $0.065
- Fits in one request
- Grok 4.20
- $0.118
- Fits in one request
Claude Haiku 4.5 has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Claude Haiku 4.5
- $0.09
- Does not fit in one request
- Grok 4.20
- $0.5
- Fits in one request
- Cached input priced at the published list-input rate
Claude Haiku 4.5 does not fit this workload in one request. Grok 4.20 has no published cached-input rate, so cached tokens use its listed input rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationClaude Haiku 4.5Grok 4.20
Context window
Maximum documented context; output-token limits may be lower.
Claude Haiku 4.5
Grok 4.20
2M
API model ID
Claude Haiku 4.5
claude-haiku-4-5-20251001
Claude API pricingGrok 4.20
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Claude Haiku 4.5
$0.1 per 1M cached input tokens
Claude API pricingGrok 4.20
Not published
Documented inputs
Claude Haiku 4.5
Not sourced
Grok 4.20
Not sourced
Documented outputs
Claude Haiku 4.5
Not sourced
Grok 4.20
Not sourced
Provider availability
Claude Haiku 4.5
Not sourced
Grok 4.20
Not sourced
Reasoning profile
Claude Haiku 4.5
Non-Reasoning
Grok 4.20
Reasoning
Weight access
Claude Haiku 4.5
Proprietary
Grok 4.20
Proprietary
License
Claude Haiku 4.5
Proprietary
Grok 4.20
Proprietary
Release date
Claude Haiku 4.5
2025-10-15
Grok 4.20
2026-03-10
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- Grok 4.20 has the higher public score estimate, 67.13 versus 52.79, but the 90% score intervals overlap.
- Workload cost
- Repository review: $0.065 vs $0.118. Cache-heavy agent loop: $0.09 vs $0.5.
- Context tradeoff
- Grok 4.20 has the larger documented window (2M).
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence27 rows
Agentic
JobBench
Claude Haiku 4.516.0%SourceGrok 4.20—Not directly comparable
Terminal-Bench 2.1 (Vals)
Claude Haiku 4.543.8%SourceGrok 4.2044.2%SourceGrok 4.20 leads this result
Terminal-Bench 2.0
Claude Haiku 4.5—Grok 4.2047.1%SourceNot directly comparable
DeepSearchQA
Claude Haiku 4.5—Grok 4.2062.8%SourceNot directly comparable
Gert Labs
Claude Haiku 4.5—Grok 4.2038.36%SourceNot directly comparable
Coding
SWE-bench Verified
Claude Haiku 4.573.3%SourceGrok 4.2076.7%SourceGrok 4.20 leads this result
VulcanBench v3
Claude Haiku 4.576.2%SourceGrok 4.20—Not directly comparable
LiveCodeBench (Vals)
Claude Haiku 4.541.2%SourceGrok 4.2084.3%SourceGrok 4.20 leads this result
SWE-bench (Vals)
Claude Haiku 4.566.6%SourceGrok 4.2072.2%SourceGrok 4.20 leads this result
LiveCodeBench Pro
Claude Haiku 4.5—Grok 4.2074.2%SourceNot directly comparable
SWE-bench Pro
Claude Haiku 4.5—Grok 4.2051.8%SourceNot directly comparable
Vibe Code Bench
Claude Haiku 4.5—Grok 4.204.06%SourceNot directly comparable
Reasoning
Knowledge
GPQA Diamond (Vals)
Claude Haiku 4.572.2%SourceGrok 4.2088.6%SourceGrok 4.20 leads this result
MMLU-Pro (Vals)
Claude Haiku 4.578.7%SourceGrok 4.2086.3%SourceGrok 4.20 leads this result
GPQA-D
Claude Haiku 4.5—Grok 4.2088.5%SourceNot directly comparable
HLE w/o tools
Claude Haiku 4.5—Grok 4.2031.6%SourceNot directly comparable
HealthBench Hard
Claude Haiku 4.5—Grok 4.2020.3%SourceNot directly comparable
MedXpertQA (Text)
Claude Haiku 4.5—Grok 4.2050.2%SourceNot directly comparable
Math
Multimodal
MMMU-Pro
Claude Haiku 4.5—Grok 4.2075.2%SourceNot directly comparable
CharXiv
Claude Haiku 4.5—Grok 4.2060.9%SourceNot directly comparable
ERQA
Claude Haiku 4.5—Grok 4.2054.1%SourceNot directly comparable
SimpleVQA
Claude Haiku 4.5—Grok 4.2057.4%SourceNot directly comparable
MedXpertQA (MM)
Claude Haiku 4.5—Grok 4.2065.8%SourceNot directly comparable
Frequently asked questions
Which is better, Claude Haiku 4.5 or Grok 4.20?
Grok 4.20 has the higher public score estimate, 67.13 versus 52.79, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, Claude Haiku 4.5 or Grok 4.20?
Grok 4.20 leads the public coding lane, 28.2 to 27.2, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, Claude Haiku 4.5 or Grok 4.20?
The like-for-like agentic tasks row is a practical tie on the public lane, 27 against 26.7, inside the 0.5-point band BenchLM treats as level.
Which costs less, Claude Haiku 4.5 or Grok 4.20?
For the stated presets, chat costs $0.0035 on Claude Haiku 4.5 and $0.005 on Grok 4.20; repository review costs $0.065 and $0.118; the cache-heavy agent loop costs $0.09 and $0.5. Claude Haiku 4.5 does not fit this workload in one request. Grok 4.20 has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, Claude Haiku 4.5 or Grok 4.20?
Grok 4.20 has the larger documented context window: 2M, compared with 200K.
Related comparisons
Last updated September 10, 2026
Watch Claude Haiku 4.5 vs Grok 4.20
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.